Page 1 of 2 1 2 LastLast
Results 1 to 15 of 21

Thread: Scores for Yahoo! Top 100 Skaters (2012-2013)

  1. #1
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Demi-God

    Default Scores for Yahoo! Top 100 Skaters (2012-2013)

    I have downloaded the Yahoo! skater data from one of my leagues: Rocky Horror Roto Show.

    It contains stats for: G, A, +/-, PIM, PPP, SOG, FOW, HIT, BLK.

    The Yahoo! "Rank" is based only on the first six.
    This is helpful to know when playing in a league like this... since a player has even more value in the three non-counted categories.

    If two players are ranked close to each other, a manager should be drafting the guy with the better statistics in FOW/HIT/BLK - since the ranking suggests the players are close in the main six cats.


    Anyways, there is some combination of numbers that sort the players into their Yahoo! ranks.

    The ranks is sorted by a sum of the following:
    G x Goal-Factor
    A x Assist-Factor
    +/- x +/--Factor
    PIM x PIM-Factor
    PPP x PPP-Factor
    SOG x SOG-Factor
    [It's also possible that it's not factors... could be to the power of 2, square, or something more intense... but I'd guess not.]

    Players that have not played (0.00 value) are ranked around the 990s.
    Just below that 0.00 value are players that are -1... with 2 SOG (Andrew Gordon).
    Just above that 0.00 value are players that are -1... with 4 SOG (Frank Corrado).

    This sort of information tells me that a SOG is worth more that 0.25... but less than 0.50... when compared to a +/- of ONE.


    Anyways - I have the entire statistical database in Excel.
    So... if anybody wants to mess with the numbers, let me know.
    I'd love to have some more MATHIES figure this thing out.

    The goalie scoring is even more difficult... and I intend on trying to conquer that one only after I work out the scorers.


    If anybody wants a copy of the Excel spreadsheet & stats, send me a PM with your email address.
    I can email it to you and we can crunch the numbers out together.

  2. #2
    Location
    Scotland
    Rep Power
    50

    The Wolverine

    Default

    I have nowhere near the maths skills but should be an interesting exercise to whomever takes it on.

    I also wonder if it's possible to include the ease and availability of acquiring such stats e.g. easier to find Hits than PPP on FA? Would that affect the value or 'rank' of specific players?

  3. #3
    Location
    TERRACE is hockey
    Rep Power
    50

    Dobber Sports Giant

    Default

    you have me confused but id love to read more into it when its done. id like to see some with a number of different league settings after its been figured out.
    somebody has there work cutout for them. lol.
    10 team Full Keeper Roto League
    4C-4LW-4RW-6D-2G-5bench-unlimited farm.

    C= eichel,horvat,malkin,kuznetsov
    LW= forsberg,kreider,lee,b.tkchuch,hertle,granlund
    RW= reinhart,radulov,hoffman,laine,hayes
    D= doughty,ghostebehere,karlsson,jones,klefbom,dahlin ,skjei
    G= price,murray,grubauer,georgiev,ullmark,samsonov

    under250gp=nichushkin,tuch,mathesson,marino,pujuja rvi,hosang,terry,andersson,j.hughes,patrik,crouse, zadina,podkolzin,k.miller,nedeljkovic,kravtsov,hay ton

    --G,A,P,PPP,SHP,GWG,PIMS,+/-,SOG,hits,FOW,blks-Wins,GAA,SV%,saves,SO

  4. #4
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Demi-God

    Default

    See attached PDF.

    Yahoo calculates a player's "rank" by using only 6 categories: G, A, +/-, PIM, PPP, SOG. On the attachment, these are the cells that are yellow.

    If somebody were to look at the players ranked just above, or just below, the players that didn't play at all (0.00 value)... you'd be able to start to get an idea how Yahoo ranks a player.

    For example:

    Frank Corrado (VAN): 0G, 0A, -1, 0 PIM, 0 PPP, 4 SOG. He has some positive value (slightly above 0.00), since Yahoo ranks him above guys that didn't play.

    Andrew Gordon (VAN): 0G, 0A, -1, 0 PIM, 0 PPP, 2 SOG. He has some negative value (slightly below 0.00), since Yahoo ranks him below guys that didn't play this season.

    So... SOG is multiplied by something in the scoring, relative to a "whole" +/- number of, +1 (say) or -1 (say).


    -1 + 4a > 0 Therefore, a > 0.25 for this to be true. (A deduction extracted from Corrado's "rank")
    -1 + 2a < 0 Therefore, a < 0.50 for this to be true. (A deduction extracted from Gordon's "rank")

    For now, I've chosen a value of a = 0.275.
    So, in my effort to mimic Yahoo scoring, I multiply SOG x 0.275.

    Other categories have similar weighting... some multiplier... I'm just trying to figure it out.

    I have a few numbers plugged in (pink)... but my scoring isn't perfect because some of my players scores aren't ordered yet:

    Example:
    Y08! Toews (Score = 153.195) should have higher score than
    Y10! Kessel (Score = 153.905)... since Yahoo's rank of Toews is higher.
    [and later... goalie algorithm scoring can be analyzed to see how Yahoo! decides that some goalie should be rated between Toews & Kessel]

    I can consistently tweak the multipliers (pink)... but at some point, it becomes very laborious.
    And I really need some input - or somebody that understands how to use Microsoft Excel's higher (solving) powers.


    Once this is solved, the spreadsheet can be used to input predictions for players in future years and determine where they should be drafted relative to somebody else.
    The spreadsheet can also be used to do the math (accurately) on any combination of categories used in different pools - valuable to us.

    This sort of math/problem-solving is something I really enjoy.
    I'm just looking for a like-minded person to undertake it with me.
    Attached Files Attached Files
    Last edited by Pengwin7; April 30, 2013 at 5:40 PM.

  5. #5
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Demi-God

    Default

    Another reason to extract "scoring" from "ranks".

    People often ask "What is the best spot to draft from? There are 12 teams in my league."

    The answer is almost always 1st.
    This has to do with Gaussian distribution of statistics (complicated).

    In the model I'm working on, these are the scores for players from this season:
    (1) 201.1 Ovechkin, Alex
    (2) 173.2 Kunitz, Chris
    (3) 168.1 Crosby, Sidney
    (4) 162.1 Subban, PK
    (5) 160.7 Stamkos, Steven
    (6) 157.8 Zetterberg, Henrik
    (7) 155.6 Kane, Patrick
    (8) 154.0 Toews, Jonathan
    (9) 154 *Some Goalie
    (10) 153.9 Kessel, Phil
    (11) 151.9 Staal, Eric
    (12) 150 *Some Goalie

    (13) 149.3 Hall, Taylor

    (14) 148.8 St.Louis, Martin
    (15) 147.3 Nash, Rich
    (16) 147.5 Datsyuk, Pavel
    (17) 146.3 Getzlaf, Ryan
    (18) 144.9 Semin, Alexander
    (19) 144 *Some Goalie
    (20) 143.2 Tavares, John
    (21) 143.1 *Some Goalie
    (22) 143.0 Ribeiro, Mike
    (23) 142 *Some Goalie
    (24) 141 *Some Goalie

    With this sort of information, you could clearly see that having the #1 & #24 picks would get you 342pts (201 + 141).
    Next best would be the #2 & #23 picks giving 315pts (173 + 142).
    The end of the 1st round, early 2nd round would be picks 12/13, 11/14, 10/15, 9/16... most of these pick combinations would add up to 300pts.

    This is one example on the value of calculating scoring to separate ranks.

  6. #6
    Rep Power
    0

    Dobber Sports Recruit

    Default

    send this info to [email protected] and he will look into this for you. paulieb

  7. #7
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Demi-God

    Default

    *** I DID IT ****


    I did it with several iterations, moving category-by-category.
    But I did figure out how to generate a "score" that sorted the Yahoo! ranks.

    Attached are skaters that finished in Yahoo!s top 100, by rank.

    This an especially nifty spreadsheet because it shows you, for example:

    *That the #1 overall player Alexander Ovechkin (198.06pts) was twice as valuable as the #98 overall player, Jason Pominville (97.62pts).

    Yahoo!s "rank" is based on full season statistics.
    But for players playing a full season - one could use this spreadsheet to determine player values in trades or comparison (similar to FHG).

    It can also be useful to input a player's projects for a season and then figure out where they rank... or should be drafted.


    My next battle:
    Figuring out how to slide in Yahoos! #1 goalie, Sergei Bobrovsky into his #9 fantasy ranking, between Toews & Kessel.

    152.88 pts #8. Jonathan Toews
    15x.xx pts #9. Sergei Bobrovsky
    151.65 pts #10. Phil Kessel

    2012-2013 Full Season Stats [G, A, +/-, PIM, PPP, SOG]
    Scores for Skaters - matching Yahoo "Rank"
    198.06 1 Alex Ovechkin
    171.55 2 Chris Kunitz
    166.39 3 Sidney Crosby
    159.01 4 P.K. Subban
    158.88 5 Steven Stamkos
    155.26 6 Henrik Zetterberg
    153.92 7 Patrick Kane
    152.88 8 Jonathan Toews
    ***This is where Yahoo slides in Sergei Bobrovsky #9... somehow.***
    151.65 10 Phil Kessel
    151.23 11 Eric Staal
    148.00 13 Taylor Hall
    146.73 14 Martin St. Louis
    146.52 15 Rick Nash
    145.89 16 Pavel Datsyuk
    144.75 17 Ryan Getzlaf
    144.19 18 Alexander Semin
    141.55 20 John Tavares
    139.81 22 Mike Ribeiro
    136.71 25 Pascal Dupuis
    136.06 27 Corey Perry
    135.21 28 Claude Giroux
    134.80 29 Max Pacioretty
    134.73 30 Derek Stepan
    131.88 31 Jakub Voracek
    131.18 33 Zach Parise
    130.41 34 Evander Kane
    128.52 35 James Neal
    128.49 36 Daniel Sedin
    127.75 37 Nicklas Backstrom
    127.12 38 Tyler Seguin
    126.93 39 Anze Kopitar
    126.20 40 Henrik Sedin
    125.17 41 Nazem Kadri
    123.61 42 Matt Moulson
    123.55 43 Joe Thornton
    122.87 44 Andrew Ladd
    122.52 45 Wayne Simmonds
    121.52 46 Logan Couture
    121.46 47 Johan Franzen
    120.59 48 Jiri Tlusty
    118.81 49 David Clarkson
    118.15 50 Thomas Vanek
    117.36 53 Evgeni Malkin
    117.32 54 Tomas Plekanec
    116.48 56 Patrik Elias
    115.96 57 Mikko Koivu
    115.52 58 Justin Williams
    115.39 59 Dustin Brown
    115.17 60 Keith Yandle
    114.05 61 Brad Marchand
    114.01 62 Patrick Marleau
    113.82 63 Kris Letang
    113.61 64 Patrice Bergeron
    111.39 66 Alexandre Burrows
    110.95 67 Michael Ryder
    110.87 68 P.A. Parenteau
    110.83 69 Marian Hossa
    110.28 70 Chris Stewart
    110.14 71 Blake Wheeler
    109.01 72 Ryan Callahan
    108.49 73 Sam Gagner
    108.01 74 Troy Brouwer
    107.50 75 Jeff Carter
    107.26 76 Jarome Iginla
    106.45 77 Dion Phaneuf
    105.83 78 Shea Weber
    103.84 79 Zdeno Chara
    103.29 80 Jordan Eberle
    103.28 81 Jaromir Jagr
    103.20 82 Dustin Byfuglien
    102.87 83 Brad Richards
    102.28 84 Joe Pavelski
    102.12 85 Duncan Keith
    101.73 86 Kimmo Timonen
    101.19 87 Matt Duchene
    100.96 88 Kyle Turris
    100.08 89 Ryan Suter
    99.66 90 Brendan Gallagher
    98.89 92 David Backes
    98.34 94 Ilya Kovalchuk
    97.69 96 Radim Vrbata
    97.62 97 Teddy Purcell
    96.96 98 Jason Pominville
    96.89 99 Shane Doan
    96.80 100 Mike Richards
    Attached Files Attached Files
    Last edited by Pengwin7; May 1, 2013 at 1:05 PM.

  8. #8
    Location
    Scotland
    Rep Power
    50

    The Wolverine

    Default

    Good work!

  9. #9
    Rep Power
    0

    Dobber Sports Prodigy

    Default

    so what if i do
    166.39/36=4.622 fantasy points [per a game- crosby
    198.06/48=4.126 fppg -ovy

    so if all the teams only drafted the top PLAYERS what would the first waiver wire/free agents fppg that would replace crosby? ie the 193rd ranked player this year? (or if center only maybe the 30th-37th ranked center but sometimes u snipe the BAB player) now do you think they included that players(x) fppg in crosby's (y) point's total? i.e.
    ypoints=(yfppg*ygp) + (xfppg*(totalgames-ygp)
    probably too many brackets

  10. #10
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Demi-God

    Default

    Quote Originally Posted by Jammin1390 View Post
    so what if i do
    166.39/36=4.622 fantasy points [per a game- crosby
    198.06/48=4.126 fppg -ovy

    so if all the teams only drafted the top PLAYERS what would the first waiver wire/free agents fppg that would replace crosby? ie the 193rd ranked player this year? (or if center only maybe the 30th-37th ranked center but sometimes u snipe the BAB player) now do you think they included that players(x) fppg in crosby's (y) point's total? i.e.
    ypoints=(yfppg*ygp) + (xfppg*(totalgames-ygp)
    probably too many brackets

    Yup - good thinking... and it's something all poolies have to account for.
    It depends on depth of the league rosters and WHO, exactly, is that best available free agent.

    Most leagues are 12 teams x 12F (9F starting, 3F bench).
    So that's 144 forwards.

    If Crosby plays 75% of the year, that's the 166.39pts he got this year.
    The 145th best forward will put up about 80pts (full year), or 20pts in that 25% (quarter year) where he is used to replace Crosby.

    So... 188.39pts, Crosby (3/4 year) + WaiverWirePlayer (1/4 year).
    Still not as good as Ovechkin (full year), but pretty close.


    GREAT EXAMPLE though, on how to gain value out of the "scores".

  11. #11
    Location
    Scotland
    Rep Power
    50

    The Wolverine

    Default

    Yep context definitely helps apply the rankings/scores. Furthermore, you also have to look at your team need and perhaps the best or 'highest score' FA may not be the best option to replace an injured player (e.g. Jammin's example of centre, FOW could be a big need etc.). So, even if the score of Player A 75% + Player B 25% < Player C 100%, their specific category strength could help you more. Or if Player A 75% + Player B 25% > Player C 100%, the extra boost might not help you if it's already a team strength.

    Like FHG, this gives you a snapshot of the rankings to see how players compare each other by looking at all the Yahoo standard categories together. But context/team needs always plays a greater role in how you apply it, especially as you (Pengwin) alluded to at the beginning when you add in the peripheral categories. Or if you consider categories that are more difficult/easier found on FA. For example, Dupuis was solid this year, but if you needed PPP, his value is decreased significantly.

  12. #12
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Demi-God

    Default

    Great post, hf.

    To further expand on context... a manager has to zig more than other managers zig.
    If people in your league are not loading up on PIM (say they roster 400 PIM), then a manager should realize that he only needs 450 PIM or 500 PIM... but does not need 800 PIM.

    The amount a category is weighted can be adjusted.
    Yahoo! has their set amount... but the value in each league changes based on how much of a category is rostered by the league.

    For example, if nobody rosters PIM goons, then a team may only need to draft Corey Perry to lock-up PIM in any match-up (or yearly standings, if ROTO).


    One thing that is nice about having a score-based spreadsheet with multipliers is that the author can adjust these multipliers to suit the league.

    For example, what I do is total up the quantity in a category that is drafted.
    Then divide by the number of skaters.

    So if the league rosters 150 players:

    If those 150 players combine for 3000 goals, that's an AVERAGE of 20 goals per player.
    I'd weight goals such that a player that scores 20 goals is worth 5/10 points.
    A player that scores 0 goals is (obviously) worth 0.
    A player that scores 40 goals is worth 10/10.

    0g player = 0pts
    20g player = 5pts
    40g player = 10pts

    If those 150 players combine for 4500 PIM, that's 30 PIM per player.
    I'd weight PIM such that a player that gets 30 PIM is worth 5/10 points.
    A player that gets 0 PIM is (obviously) worth 0.
    A player that gets 60 PIM is worth 10/10 points... AND...
    a player that gets 120 PIM is worth... (real-value)... 20points!.

    0 PIM player = 0pts
    30 PIM player = 5pts
    60 PIM player = 10pts
    120 PIM player = 20pts

    The players get a score based on their totals across categories.
    You draft based on the best "score"... but you do have to keep in mind while you draft:

    ****
    My tip to anybody making their own "scoring" spreadsheet.
    In quantity categories, a stat count of 0 (0 G, 0A... 0 SOG), should be worth a category score of 0.0. If you don't roster a player, it's 0 stats.
    In quantity categories, a stat count of "average", should be worth a category score of 5.0. It can really be any number... but I like 5, since typically we think of a 10/10 player as the upper end limit... elite.

    One thing I like about my "scoring" spreadsheet is that the "scoring system" works off of +1 or -1. +1 = +1 scoring point. It's the basis of my scoring.
    So, if somebody wanted to quickly say "What if Pascal Dupuis finished +0 instead of +30?"... well, it's as easy as taking 30pts of Dupuis' score.

    In the past year, the best example of "value" I saw in this exercise was Eric Staal.
    E.Staal finished at -20 last year... but with new linemate Semin (& Tlusty... and possibly brother Jordan), we all knew his +/- should correct.
    But... what is that value? How much higher do we rank him?
    If we have "scores"... we know... we can do that exercise.
    Same thing with Rick Nash.
    He goes from Columbus, where he's -20 usually, to NYR, where he'll certainly be PLUS.
    How much does his "rank" change? We don't know.
    How much does his "score" change? Well, we can figure that out... and once we do that, we'll know his new rank.

    BOTH Eric Staal and Rick Nash were guys I drafted this year in my RHRS Roto league.
    Major contributors to my team performance, BOTH being drafted by me outside of the top 15 picks and BOTH finishing as top 15 Yahoo fantasy players
    ****

    i) Though you can pick the "highest score"... one has to consider how the value of that player's score is developped.
    A mistake I saw in several leagues this year is people selecting a guy like Steve Ott too early.
    High "score", yes.... but could those categories for which the high score is generated be drafted later... or picked up from the waiver wire??? For a guy like Ott, whose value is Hits/PIM/FOW, the answer is YES.
    ii) Which categories can be plucked off the wire during the season (typically, Assists, +/-, PIM, SOG, HITS, BLK, FOW)
    iii) Which categories can not be plucked off the wire during the season (typically Goals & PPP are hard to find)


    The one thing my number crunching taught me this year is this:
    *The category that one MUST load up on during the front end of the draft is... PPP.
    This is the hardest category to get later in the season. PP1 players rarely emerge during the season.
    This is why elite skaters (forwards) and #1 defensemen should be drafted first.

    Lots of other value in seeing "scores" instead of "ranks"... too.
    Enough for one post....
    Last edited by Pengwin7; May 3, 2013 at 10:24 AM.

  13. #13
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Demi-God

    Default

    One more example of understanding "scoring" and "category score".

    Say a league has 5 categories.
    G, A, PIM, W, SV%.

    You, a buddy, and another buddy get 25.0 "coins" to buy player stats.
    You can buy a score in a category of 0-10, each point in a category costs one "coin".

    Buddy A invests: 5.0 (G), 5.0 (A), 5.0 (PIM), 5.0 (W), 5.0 (SV%)
    Buddy B invests: 9.0 (G), 8.0 (A), 8.0 (PIM), 0.0 (W), 0.0 (SV%)
    Buddy C invests: 0.0 (G), 9.0 (A), 9.0 (PIM), 6.0 (W), 1.0 (SV%)

    Who has the best team in H2H?
    Who has the best team in Roto?

    If you can figure these things out... then having a "scoring" spreadsheet with "category scores" is the THING FOR YOU!!!

  14. #14
    Rep Power
    0

    Dobber Sports Prodigy

    Default

    Quote Originally Posted by Pengwin7 View Post
    Great post, hf.

    To further expand on context... a manager has to zig more than other managers zig.

    If people in your league are not loading up on PIM (say they roster 400 PIM), then a manager should realize that he only needs 450 PIM or 500 PIM... but does not need 800 PIM.

    This is the hardest category to get later in the season. PP1 players rarely emerge during the season.
    This is why elite skaters (forwards) and #1 defense men should be drafted first.
    In quantity categories, a stat count of 0 (0 G, 0A... 0 SOG), should be worth a category score of 0.0. If you don't roster a player, it's 0 stats.

    .
    I prefer the zag of the h2h format, easier to rotate for the needed cats to win. So to win every week, you want to just win.So to obtain a balanced roster why not compare it to a stat count of 0 being the avg baseline of the 140-148th or whatever ranked player in the league for a few years instead of a 0.0.0.0.0 baseline. Because it will show which stat is the easiest to grab in free agency thus the one that should be less favoured in the draft?

  15. #15
    Location
    Prairies
    Rep Power
    40

    Dobber Sports Demi-God

    Default

    Quote Originally Posted by Jammin1390 View Post
    why not compare it to a stat count of 0 being the avg baseline of the 140-148th or whatever ranked player in the league for a few years instead of a 0.0.0.0.0 baseline.
    LOL.
    Because every league size changes.

    Relative difference (subtraction) between two positions is always the same.

    If I were to set a 0.0 at some rank (140th), then larger leagues would be rostering players with a perceived negative value.
    Nobody wants to be subtracting a negative number from a positive number.
    That's just nutty.

    A rostered player is always better than 0 stats (unless he's -1 with no stats).
    That's the correct comparison.

    I've had this discussion with some of the esteemed mathematicians on these forums (and... passively, won my argument - since they eventually changed their algorithms and I have not).

    Mathematicians & scientists should work with absolute zero.
    Kelvins, my friend... I'm rockin' the Kelvins.
    Last edited by Pengwin7; May 3, 2013 at 2:36 PM.

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •